Data Analyst

EKATERINA SOTIRIS

mobile: 571-***-****

Contact

********@****.***.***

Information

University of Maryland, College Park, MD 8/2013

Education

Ph.D., Mathematical Statistics

Thesis topic: Outlier detection in Gaussian spatial random fields

Advisors: Professor Eric V. Slud, Dr. William R. Bell

Columbia University, New York, NY 5/2003

M.A., Mathematics of Finance

University of Minnesota, Minneapolis, MN 5/2002

B.S., Mathematics and Statistics

Robert J. Buehler Scholarship, Dean’s List

6/2004 7/2006

Economic Systems, Inc., Washington DC

Professional

Research Analyst

Experience

Performed analysis of customer loyalty and profitability for Bank of America.

Addressed lifetime profitability of bank’s customers, which involved analyzing

longitudinal customer survey data. Analyzed complex datasets, utilizing methods

such as linear and logistic regression and survival modeling.

Conducted research for the Florida Department of Health on patient and employee

satisfaction, based on data from a yearly satisfaction survey. Used multivariate

data analysis techniques to identify drivers of customer satisfaction.

Analyzed Census and home mortgage disclosure data for the Department of

Housing and Urban Development in a study aimed at redefining underserved

areas in non-metropolitan locations. Roles included developing the appropriate

statistical methodology for data analysis and defining a sampling scheme for

surveying households.

11/2003 5/2004

LaBranche Structured Products, New York NY

Quantitative Analyst

Used Visual Basic to develop automated and customized Excel tools to support

the company’s trading strategies.

Researched statistical methodology related to derivatives trading with the goal

of enhancing the company’s trading strategy.

6/2003 11/2003

ACNielsen, Paramus NJ

Research Analyst

Applied multivariate analysis techniques to answer questions related to product

placement in major national grocery stores. Created PowerPoint presentations

for client review. Interfaced with major clients, such as Nabisco.

2/2009 6/2011

U.S. Census Bureau, Suitland MD

Academic

Doctoral Researcher

Appointments

Conducted a study to evaluate the short-term forecast accuracy of non-seasonal

movements in economic time-series data. Examined how forecast accuracy of

the non-seasonal component is affected by the following choices: (1) alternative

choices of simple models for forecasting the seasonally adjusted series; (2) the

use of seasonally adjusted vs. trend estimates; (3) use of revised vs. unrevised

seasonally adjusted data; (4) use of X-11 vs. ARIMA model-based adjustment;

(5) use of seasonally adjusted data in forecasting vs. directly forecasting the

unadjusted series.

1 of 2

Investigated the Gaussian Mixture (GM) modeling approach to outlier detection

in spatial random fields. Derived and studied a statistic for outlier detection

based on the posterior probability of outliers under the GM model. Studied the

applicability of the EM algorithm in obtaining maximum likelihood parameter

estimates in spatially correlated GM models.

Worked towards extending existing time series outlier detection methodology to

higher dimensional spatial data. Developed methodology for setting hypothesis

test thresholds.

9/2008 12/2008

Instructor of Mathematics

Department of Mathematics, University of Maryland, College Park MD

Taught Introduction to Probability as a sole contact instructor.

Oversaw all aspects of the course, including course and exam preparation, teaching,

grading, office hours and recommendation letter preparation.

8/2006 12/2008

Teaching Assistant

Department of Mathematics, University of Maryland, College Park MD

Taught recitations for Calculus II and Linear Algebra, held office hours, prepared

and graded quizzes and tests.

Received many outstanding teaching evaluations and recognition.

Conferences & Sotiris E., Bell W., Seasonal Adjustment to Facilitate Forecasting: Empirical Results.

In: Proceedings of the Joint Statistical Meetings, August 2010, Vancouver CA.

Talks

Sotiris E., Spatial Modeling using a new class of non-stationary covariance functions.

In: Time Series Division Seminar, January 2010, Suitland MD.

Sotiris E., Forecasting an accumulated series based on partial accumulation: a Baysian

Method for short series with seasonal pattern. In: Time Series Division Seminar,

March 2009, Suitland MD

Sotiris E., Outlier Detection in Gaussian Spatial Random Fields, ProQuest 2013

Thesis

Extended popular time series outlier detection methodology to higher-dimensional

(spatial) data. Developed methodology for setting hypothesis test thresholds utilizing

theory on the maxima of discretely sampled Gaussian random fields. Studied the

effect of undetected outliers on spatial prediction. Considered asymptotic properties

of maximum likelihood parameter estimates in Gaussian random fields.

SAS, R, VBA, SQL

Languages

Dr. Eric Slud (email: ***@****.***.***; phone: 301-***-****)

References

Professor, Department of Mathematics, University of Maryland, College Park MD

Available to

Dr. William Bell (email: *******.*.****@******.***; phone: 301-***-****)

Contact

Research and Methodology Directorate, U.S. Census Bureau, Washington DC

Dr. George Kettner (email: ********@*******.***; phone: 703-***-****)

President, Economic Systems, Inc., Falls Church VA

2 of 2

Contact this candidate